[openstack-dev] [ironic] 100% failure in pxe_ssh job

Sean Dague sean at dague.net
Tue Sep 1 13:27:53 UTC 2015


On 09/01/2015 09:21 AM, Julia Kreger wrote:
> 
> On Tue, Sep 1, 2015 at 9:17 AM, Jim Rollenhagen <jim at jimrollenhagen.com
> <mailto:jim at jimrollenhagen.com>> wrote:
<snip>
> 
>     Why don't we just use the pre-built agent ramdisks we already
>     publish? AFAIK we already have a job for it, just need to switch the
>     name in project-config for Nova.
> 
>     // jim
> 
> 
> We need to make sure the dib element is in working shape, which means
> means it actually has to be utilized somewhere along the line.

Which should be a functional test on dib, or at some lower level, not
randomly done in the integration tests of unrelated projects. The
current ball of mud testing leads to high coupling and high failure
rates, because everything is secondarily tested. It should never be run
during nova patches, because there is 0 chance that a Nova patch can
impact that code.

We spent a lot of time trying to decouple libraries from projects and
put the tests in the right places to actually verify the behavior closer
to the code in question. dib / ironic is still doing gating like other
projects were a year ago, which was fragile and led to unrelated lock
ups like this really often. So that needs to be prioritized soon.

	-Sean

-- 
Sean Dague
http://dague.net



More information about the OpenStack-dev mailing list